def test_read_table(self):
pre = PrepareData('test', '../../data/demo_data/all')
admissions = pre.read_table('admissions')
# Check data type of table
self.assertEqual(type(admissions), pd.core.frame.DataFrame,
'Read table is not of type Pandas.DataFrame.')
# Check if there is data in the table
self.assertEqual(admissions['DIAGNOSIS'][2], 'SEPSIS',
'Data not read correctly')
# Check if data is loaded correctly and if datetimes are marked correcly in table
self.assertEqual(
(admissions['DISCHTIME'] - admissions['ADMITTIME'])[3],
pd.Timedelta('8 days 01:23:00'),
'Timedates not correctly marked in table.')
# Test pickling files: First create empty file, use save-function of read_table, check if stored correctly
file_name_pickle = '../../data/demo_data/all/pickle/admissions.p'
empty_table = pd.DataFrame()
pickle.dump(empty_table, open(file_name_pickle, "wb"))
admissions = pre.read_table('admissions', True)
loaded_table = pickle.load(open(file_name_pickle, 'rb'))
print(loaded_table)
self.assertEqual(loaded_table['DIAGNOSIS'][2], 'SEPSIS',
'Data is not correctly pickled.')
def test_get_table_names(self):
pre = PrepareData('test', '../../data/demo_data/all')
tables_in_folder = pre.get_table_names(True)
self.assertEqual(tables_in_folder[0], 'admissions',
'Does not return tables names correctly.')
tables_in_folder = pre.get_table_names(False)
self.assertEqual(tables_in_folder[0], 'ADMISSIONS',
'Does not return tables names correctly.')
def test_read_tables(self):
pre = PrepareData('test', '../../data/demo_data/all')
table_list = ['admissions', 'patients']
pre.read_tables(table_list)
# Check if the two tables are loaded
self.assertEqual(pre.admissions['DIAGNOSIS'][2], 'SEPSIS',
'Not read multiple tables correctly.')
self.assertEqual(pre.patients['SUBJECT_ID'][3], int(10017),
'Not read multiple tables correctly.')
def test__get_drug_feature(self):
pre = PrepareData('test', '../../data/demo_data/all')
pre.read_table('prescriptions')
drug_enc = DrugEncoderV2(48, None)
drug_enc.fit(pre.prescriptions)
unique_drugs = drug_enc._get_drug_feature(pre.prescriptions)
self.assertEqual(
len(unique_drugs), len(set(unique_drugs)),
'Error in DrugEncocder._get_drug_feature(); list elements are not unique;'
)
def test_load_table_pickle(self):
pre_store = PrepareData('test1', '../../data/demo_data/all')
pre_store.read_table('admissions', True)
pre_load = PrepareData('test2', '../../data/demo_data/all')
pre_load.load_table_pickle('admissions')
self.assertEqual(pre_load.admissions['DIAGNOSIS'][2], 'SEPSIS',
'Not loaded table from pickle file.')
def test_load_table_pickle_list(self):
table_list = ['admissions', 'patients']
pre_store = PrepareData('test1', '../../data/demo_data/all')
pre_store.read_tables(table_list, True)
pre_load = PrepareData('test2', '../../data/demo_data/all')
pre_load.load_table_pickle_list(table_list)
self.assertEqual(pre_load.admissions['DIAGNOSIS'][2], 'SEPSIS',
'Not loaded multipe tables from pickle files.')
self.assertEqual(pre_load.patients['SUBJECT_ID'][3], int(10017),
'Not loaded multipe tables from pickle files.')
def test_transform(self):
pre = PrepareData('test', '../../data/demo_data/all')
pre.read_tables(['prescriptions', 'admissions'])
drug_enc = DrugEncoderV2(48, 10)
drug_enc.fit(pre.prescriptions)
hadm_drug_matrix, hadm_id_mapping = drug_enc.transform(
pre.prescriptions, pre.admissions)
self.assertTrue(
((hadm_drug_matrix[7, :] == [0, 1, 1, 1, 0, 1, 1, 1, 1, 0]) *
1).sum(),
'Error in DrugEncoder.transform(); hadm_drug_matrix wrong')
self.assertEqual(
hadm_drug_matrix.shape, (len(pre.admissions), 10),
'Error in DrugEncoder.transform(); hadm_drug_matrix has wrong shape;'
)
self.assertEqual(len(hadm_id_mapping),
pre.admissions['HADM_ID'].unique().shape[0],
'Error in DrugEncoder.transform(); wrong shape;')
self.assertEqual(
drug_enc.drug_mapping, {
'FURO40I': 0,
'NS500': 1,
'NS1000': 2,
'NACLFLUSH': 3,
'INSULIN': 4,
'D5W250': 5,
'VANC1F': 6,
'VANCOBASE': 7,
'HEPA5I': 8,
'KCL20PM': 9
}, 'Error in DrugEncoder.transform(); drug_mapping is wrong')
def test_fit(self):
pre = PrepareData('test', '../../data/demo_data/all')
pre.read_table('prescriptions')
drug_enc = DrugEncoderV2()
drug_enc.fit(pre.prescriptions)
self.assertEqual(drug_enc.drug_mapping['NS500'], int(1),
'Drug mapping not correct.')
self.assertEqual(drug_enc.drug_mapping['MORPH100'], int(797),
'Drug mapping not correct.')
# Test Encoder Parameters
pre = PrepareData('test', '../../data/demo_data/all')
pre.read_table('prescriptions')
drug_enc = DrugEncoderV2(48, 13)
drug_enc.fit(pre.prescriptions)
self.assertEqual(len(drug_enc.drug_mapping), int(13),
'max_number_of_drug_features does not work.')